文献速递腹部CT

基于ctSLEP评分与能谱CT的局部进展期胃癌新辅助治疗反应预测

Pretreatment prediction of neoadjuvant therapy response in locally advanced gastric cancer using an additive ctSLEP score and spectral CT.
PubMed 文献 · Abdom Radiol (NY) · 2026-10-31

本研究回顾性纳入158例接受新辅助治疗及根治性胃切除术的局部进展期胃癌患者,构建基于四种常规CT形态学侵犯征象的附加ctSLEP评分,并联合能谱CT定量参数及系统性炎症指标建立分层预测模型。联合模型在内验证队列中AUC最高(0.754),Brier评分最低(0.200),提示整合形态学侵犯负荷、能谱强化及系统性炎症的低维可解释策略有助于治疗前新辅助治疗反应分层。

要点

  • ctSLEP评分由浆膜侵犯、壁外静脉侵犯、隐匿性腹膜/大网膜受累细微征象及淋巴血管侵犯四项治疗前CT征象相加构成。
  • 联合模型(ctSLEP+能谱CT参数+NLR)内验证AUC为0.754,Brier评分0.200,为五个分层模型中最高。
  • 仅化疗亚组中,ctSLEP评分越高非反应 odds 越高(OR=2.15),静脉期碘浓度越高非反应 odds 越低(OR=0.83)。

关键影像表现/征象

  • ctSLEP评分由四项治疗前CT形态学侵犯征象相加构成:浆膜侵犯、壁外静脉侵犯、隐匿性腹膜/大网膜受累细微征象及淋巴血管侵犯。
  • 内验证队列中Morphology、Spectral、Imaging、Clinical-spectral及Combined模型的AUC分别为0.642、0.741、0.725、0.713和0.754。
  • 联合模型AUC最高且Brier评分最低(0.200)。
  • 形态学模型加入能谱CT后AUC提高0.083(P=0.129),Clinical-spectral模型加入ctSLEP评分后AUC提高0.041(P=0.558)。

诊断标准与关键数值

  • 以Becker肿瘤消退分级1a/1b定义为新辅助治疗反应,2/3定义为非反应。
  • 仅化疗亚组(51例)中,ctSLEP评分较高与非反应 odds 升高相关(OR=2.15,95% CI 1.35–3.88)。
  • 仅化疗亚组中,静脉期碘浓度(ICVP)较高与非反应 odds 降低相关(OR=0.83,95% CI 0.69–0.96)。
  • 内验证队列联合模型AUC为0.754,Brier评分为0.200。

临床应用与阅片要点

  • 治疗前阅片时可系统评估浆膜侵犯、壁外静脉侵犯、隐匿性腹膜/大网膜受累细微征象及淋巴血管侵犯四项征象并累加为ctSLEP评分辅助风险分层。
  • 联合能谱CT静脉期碘浓度及动脉强化分数等定量参数与中性粒细胞/淋巴细胞比值可提供增量预测信息。
  • 预测性能提升的配对比较未达统计学显著性(P=0.129及P=0.558),提示各成分的增量价值需外部验证确认。

局限性

  • 单中心回顾性研究,样本量有限(158例),结论需外部验证。
  • 能谱CT相对形态学的AUC增量及ctSLEP相对Clinical-spectral模型的AUC增量均未达统计学显著性(P=0.129、P=0.558)。
  • 仅化疗敏感性分析样本量较小(51例),亚组结果解释需谨慎。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

英文原文(摘要)· 点击展开

PURPOSE: To develop an additive ctSLEP score from routine CT morphologic invasion signs and evaluate the complementary value of morphologic CT, quantitative spectral CT, and systemic inflammation for pretreatment prediction of neoadjuvant therapy (NAT) response in locally advanced gastric cancer (LAGC).

METHODS: This retrospective study included 158 patients with LAGC who underwent NAT and radical gastrectomy and were randomly divided into training and internal validation cohorts (7:3). Becker tumor regression grade 1a/1b defined response and 2/3 nonresponse. Additive ctSLEP summed four pretreatment CT signs: serosal invasion, extramural venous invasion, subtle signs of occult peritoneal/omental involvement, and lymphovascular invasion. Five hierarchical models combined additive ctSLEP, venous-phase iodine concentration (ICVP), arterial enhancement fraction (AEF), and neutrophil-to-lymphocyte ratio (NLR). Performance was assessed using AUC, Brier score, calibration, and prespecified paired DeLong comparisons. A chemotherapy-only sensitivity analysis was performed.

RESULTS: Sixty-five patients (41.1%) were responders and ninety-three (58.9%) nonresponders. In the internal validation cohort, AUCs for the Morphology, Spectral, Imaging, Clinical-spectral, and Combined models were 0.642, 0.741, 0.725, 0.713, and 0.754, respectively. The Combined model had the highest AUC and lowest Brier score (0.200). Adding spectral CT to morphology increased AUC by 0.083 (P=0.129), and adding additive ctSLEP to the Clinical-spectral model increased AUC by 0.041 (P=0.558). In 51 patients receiving chemotherapy alone, higher additive ctSLEP was associated with greater odds of nonresponse (OR=2.15, 95% CI 1.35-3.88), whereas higher ICVP was associated with lower odds (OR=0.83, 95% CI 0.69-0.96).

CONCLUSION: The combined model showed reasonable predictive performance in internal validation. These findings support further external evaluation of a low-dimensional, clinically interpretable strategy integrating morphologic invasion burden, spectral enhancement, and systemic inflammation for pretreatment NAT response stratification.

来源:Abdom Radiol (NY),查看原文。版权归原作者所有。

引用

  1. Peng Y, Guo X, Ren T, et al. Pretreatment prediction of neoadjuvant therapy response in locally advanced gastric cancer using an additive ctSLEP score and spectral CT. Abdom Radiol (NY). 2026. PMID: 42965589. doi:10.1007/s00261-026-05855-w

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。